PreFM: Online Audio-Visual Event Parsing via Predictive Future Modeling

Audio-visual event parsing plays a crucial role in understanding multimodal video content, but existing methods typically rely on offline processing of entire videos with huge model sizes, limiting their real-time applicability. We introduce Online Audio-Visual Event Parsing (On-AVEP), a novel paradigm for parsing audio, visual, and audio-visual events by sequentially analyzing incoming video streams. The On-AVEP task necessitates models with two key capabilities: (1) Accurate online inference, to effectively distinguish events with unclear and limited context in online settings, and (2) Real-time efficiency, to balance high performance with computational constraints. To cultivate these, we propose the Predictive Future Modeling (PreFM) framework featured by (a) predictive multimodal future modeling to infer and integrate beneficial future audio-visual cues, thereby enhancing contextual understanding and (b) modality-agnostic robust representation along with focal temporal prioritization to improve precision and generalization. Extensive experiments on the UnAV-100 and LLP datasets show PreFM significantly outperforms state-of-the-art methods by a large margin with significantly fewer parameters, offering an insightful approach for real-time multimodal video understanding. Code is available at https://github.com/XiaoYu-1123/PreFM.

OmniAD: Detect and Understand Industrial Anomaly via Multimodal Reasoning

While anomaly detection has made significant progress, generating detailed analyses that incorporate industrial knowledge remains a challenge. To address this gap, we introduce OmniAD, a novel framework that unifies anomaly detection and understanding for fine-grained analysis. OmniAD is a multimodal reasoner that combines visual and textual reasoning processes. The visual reasoning provides detailed inspection by leveraging Text-as-Mask Encoding to perform anomaly detection through text generation without manually selected thresholds. Following this, Visual Guided Textual Reasoning conducts comprehensive analysis by integrating visual perception. To enhance few-shot generalization, we employ an integrated training strategy that combines supervised fine-tuning (SFT) with reinforcement learning (GRPO), incorporating three sophisticated reward functions. Experimental results demonstrate that OmniAD achieves a performance of 79.1 on the MMAD benchmark, surpassing models such as Qwen2.5-VL-7B and GPT-4o. It also shows strong results across multiple anomaly detection benchmarks. These results highlight the importance of enhancing visual perception for effective reasoning in anomaly understanding. All codes and models will be publicly available.

AlignGen: Boosting Personalized Image Generation with Cross-Modality Prior Alignment

Personalized image generation aims to integrate user-provided concepts into text-to-image models, enabling the generation of customized content based on a given prompt. Recent zero-shot approaches, particularly those leveraging diffusion transformers, incorporate reference image information through multi-modal attention mechanism. This integration allows the generated output to be influenced by both the textual prior from the prompt and the visual prior from the reference image. However, we observe that when the prompt and reference image are misaligned, the generated results exhibit a stronger bias toward the textual prior, leading to a significant loss of reference content. To address this issue, we propose AlignGen, a Cross-Modality Prior Alignment mechanism that enhances personalized image generation by: 1) introducing a learnable token to bridge the gap between the textual and visual priors, 2) incorporating a robust training strategy to ensure proper prior alignment, and 3) employing a selective cross-modal attention mask within the multi-modal attention mechanism to further align the priors. Experimental results demonstrate that AlignGen outperforms existing zero-shot methods and even surpasses popular test-time optimization approaches.

Guiding the Experts: Semantic Priors for Efficient and Focused MoE Routing

Mixture-of-Experts (MoE) models have emerged as a promising direction for scaling vision architectures efficiently. Among them, Soft MoE improves training stability by assigning each token to all experts via continuous dispatch weights. However, current designs overlook the semantic structure which is implicitly encoded in these weights, resulting in suboptimal expert routing. In this paper, we discover that dispatch weights in Soft MoE inherently exhibit segmentation-like patterns but are not explicitly aligned with semantic regions. Motivated by this observation, we propose a foreground-guided enhancement strategy. Specifically, we introduce a spatially aware auxiliary loss that encourages expert activation to align with semantic foreground regions. To further reinforce this supervision, we integrate a lightweight LayerScale mechanism that improves information flow and stabilizes optimization in skip connections. Our method necessitates only minor architectural adjustments and can be seamlessly integrated into prevailing Soft MoE frameworks. Comprehensive experiments on ImageNet-1K and multiple smaller-scale classification benchmarks not only showcase consistent performance enhancements but also reveal more interpretable expert routing mechanisms.

DiffusionReward: Enhancing Blind Face Restoration through Reward Feedback Learning

Reward Feedback Learning (ReFL) has recently shown great potential in aligning model outputs with human preferences across various generative tasks. In this work, we introduce a ReFL framework, named DiffusionReward, to the Blind Face Restoration task for the first time. DiffusionReward effectively overcomes the limitations of diffusion-based methods, which often fail to generate realistic facial details and exhibit poor identity consistency. The core of our framework is the Face Reward Model (FRM), which is trained using carefully annotated data. It provides feedback signals that play a pivotal role in steering the optimization process of the restoration network. In particular, our ReFL framework incorporates a gradient flow into the denoising process of off-the-shelf face restoration methods to guide the update of model parameters. The guiding gradient is collaboratively determined by three aspects: (i) the FRM to ensure the perceptual quality of the restored faces; (ii) a regularization term that functions as a safeguard to preserve generative diversity; and (iii) a structural consistency constraint to maintain facial fidelity. Furthermore, the FRM undergoes dynamic optimization throughout the process. It not only ensures that the restoration network stays precisely aligned with the real face manifold, but also effectively prevents reward hacking. Experiments on synthetic and wild datasets demonstrate that our method outperforms state-of-the-art methods, significantly improving identity consistency and facial details. The source codes, data, and models are available at: https://github.com/01NeuralNinja/DiffusionReward.

StabStitch++: Unsupervised Online Video Stitching with Spatiotemporal Bidirectional Warps

We retarget video stitching to an emerging issue, named warping shake, which unveils the temporal content shakes induced by sequentially unsmooth warps when extending image stitching to video stitching. Even if the input videos are stable, the stitched video can inevitably cause undesired warping shakes and affect the visual experience. To address this issue, we propose StabStitch++, a novel video stitching framework to realize spatial stitching and temporal stabilization with unsupervised learning simultaneously. First, different from existing learning-based image stitching solutions that typically warp one image to align with another, we suppose a virtual midplane between original image planes and project them onto it. Concretely, we design a differentiable bidirectional decomposition module to disentangle the homography transformation and incorporate it into our spatial warp, evenly spreading alignment burdens and projective distortions across two views. Then, inspired by camera paths in video stabilization, we derive the mathematical expression of stitching trajectories in video stitching by elaborately integrating spatial and temporal warps. Finally, a warp smoothing model is presented to produce stable stitched videos with a hybrid loss to simultaneously encourage content alignment, trajectory smoothness, and online collaboration. Compared with StabStitch that sacrifices alignment for stabilization, StabStitch++ makes no compromise and optimizes both of them simultaneously, especially in the online mode. To establish an evaluation benchmark and train the learning framework, we build a video stitching dataset with a rich diversity in camera motions and scenes. Experiments exhibit that StabStitch++ surpasses current solutions in stitching performance, robustness, and efficiency, offering compelling advancements in this field by building a real-time online video stitching system.

EvEnhancer: Empowering Effectiveness, Efficiency and Generalizability for Continuous Space-Time Video Super-Resolution with Events

Continuous space-time video super-resolution (C-STVSR) endeavors to upscale videos simultaneously at arbitrary spatial and temporal scales, which has recently garnered increasing interest. However, prevailing methods struggle to yield satisfactory videos at out-of-distribution spatial and temporal scales. On the other hand, event streams characterized by high temporal resolution and high dynamic range, exhibit compelling promise in vision tasks. This paper presents EvEnhancer, an innovative approach that marries the unique advantages of event streams to elevate effectiveness, efficiency, and generalizability for C-STVSR. Our approach hinges on two pivotal components: 1) Event-adapted synthesis capitalizes on the spatiotemporal correlations between frames and events to discern and learn long-term motion trajectories, enabling the adaptive interpolation and fusion of informative spatiotemporal features; 2) Local implicit video transformer integrates local implicit video neural function with cross-scale spatiotemporal attention to learn continuous video representations utilized to generate plausible videos at arbitrary resolutions and frame rates. Experiments show that EvEnhancer achieves superiority on synthetic and real-world datasets and preferable generalizability on out-of-distribution scales against state-of-the-art methods. Code is available at https://github.com/W-Shuoyan/EvEnhancer.

Revisiting Radar Camera Alignment by Contrastive Learning for 3D Object Detection

Recently, 3D object detection algorithms based on radar and camera fusion have shown excellent performance, setting the stage for their application in autonomous driving perception tasks. Existing methods have focused on dealing with feature misalignment caused by the domain gap between radar and camera. However, existing methods either neglect inter-modal features interaction during alignment or fail to effectively align features at the same spatial location across modalities. To alleviate the above problems, we propose a new alignment model called Radar Camera Alignment (RCAlign). Specifically, we design a Dual-Route Alignment (DRA) module based on contrastive learning to align and fuse the features between radar and camera. Moreover, considering the sparsity of radar BEV features, a Radar Feature Enhancement (RFE) module is proposed to improve the densification of radar BEV features with the knowledge distillation loss. Experiments show RCAlign achieves a new state-of-the-art on the public nuScenes benchmark in radar camera fusion for 3D Object Detection. Furthermore, the RCAlign achieves a significant performance gain (4.3\% NDS and 8.4\% mAP) in real-time 3D detection compared to the latest state-of-the-art method (RCBEVDet).

3DResT: A Strong Baseline for Semi-Supervised 3D Referring Expression Segmentation

3D Referring Expression Segmentation (3D-RES) typically requires extensive instance-level annotations, which are time-consuming and costly. Semi-supervised learning (SSL) mitigates this by using limited labeled data alongside abundant unlabeled data, improving performance while reducing annotation costs. SSL uses a teacher-student paradigm where teacher generates high-confidence-filtered pseudo-labels to guide student. However, in the context of 3D-RES, where each label corresponds to a single mask and labeled data is scarce, existing SSL methods treat high-quality pseudo-labels merely as auxiliary supervision, which limits the model's learning potential. The reliance on high-confidence thresholds for filtering often results in potentially valuable pseudo-labels being discarded, restricting the model's ability to leverage the abundant unlabeled data. Therefore, we identify two critical challenges in semi-supervised 3D-RES, namely, inefficient utilization of high-quality pseudo-labels and wastage of useful information from low-quality pseudo-labels. In this paper, we introduce the first semi-supervised learning framework for 3D-RES, presenting a robust baseline method named 3DResT. To address these challenges, we propose two novel designs called Teacher-Student Consistency-Based Sampling (TSCS) and Quality-Driven Dynamic Weighting (QDW). TSCS aids in the selection of high-quality pseudo-labels, integrating them into the labeled dataset to strengthen the labeled supervision signals. QDW preserves low-quality pseudo-labels by dynamically assigning them lower weights, allowing for the effective extraction of useful information rather than discarding them. Extensive experiments conducted on the widely used benchmark demonstrate the effectiveness of our method. Notably, with only 1% labeled data, 3DResT achieves an mIoU improvement of 8.34 points compared to the fully supervised method.

DIDS: Domain Impact-aware Data Sampling for Large Language Model Training

Large language models (LLMs) are commonly trained on multi-domain datasets, where domain sampling strategies significantly impact model performance due to varying domain importance across downstream tasks. Existing approaches for optimizing domain-level sampling strategies struggle with maintaining intra-domain consistency and accurately measuring domain impact. In this paper, we present Domain Impact-aware Data Sampling (DIDS). To ensure intra-domain consistency, a gradient clustering algorithm is proposed to group training data based on their learning effects, where a proxy language model and dimensionality reduction are employed to reduce computational overhead. To accurately measure domain impact, we develop a Fisher Information Matrix (FIM) guided metric that quantifies how domain-specific parameter updates affect the model's output distributions on downstream tasks, with theoretical guarantees. Furthermore, to determine optimal sampling ratios, DIDS combines both the FIM-guided domain impact assessment and loss learning trajectories that indicate domain-specific potential, while accounting for diminishing marginal returns. Extensive experiments demonstrate that DIDS achieves 3.4% higher average performance while maintaining comparable training efficiency.